About: The GPML toolbox is a flexible and generic Octave/Matlab implementation of inference and prediction with Gaussian process models. The toolbox offers exact inference, approximate inference for non-Gaussian likelihoods (Laplace's Method, Expectation Propagation, Variational Bayes) as well for large datasets (FITC, VFE, KISS-GP). A wide range of covariance, likelihood, mean and hyperprior functions allows to create very complex GP models.Changes:
A major code restructuring effort did take place in the current release unifying certain inference functions and allowing more flexibility in covariance function composition. We also redesigned the whole derivative computation pipeline to strongly improve the overall runtime. We finally include grid-based covariance approximations natively.
More generic sparse approximation using Power EP
Approximate covariance object unifying sparse approximations, grid-based approximations and exact covariance computations
Hiearchical structure of covariance functions
Faster derivative computations for mean and cov functions
New mean functions
New GLM link function
About: Data-efficient policy search framework using probabilistic Gaussian process modelsChanges:
Initial Announcement on mloss.org.